[Lustre-discuss] lvbo_init failed

[Charles Taylor]

Awesome.    We'll try it and let you know.  Of course, if it works,  
we'll have a naming dilemma - "The Dilger Procedure" is already  
taken.  :)

Thanks,

Charlie Taylor
UF HPC Center


On Jul 17, 2008, at 7:00 PM, Andreas Dilger wrote:

>
> This is an indication that some object is missing on the OST that one
> or more clients is trying to access.  You can look at the Lustre debug
> logs with "rpctrace" enabled to extract the "Handling RPC" and  
> "Handled RPC"
> messages on the thread printing this message, e.g.:
>
> 00000100:00100000:0:1216234365.071325:1536:32091:0:(service.c: 
> 1064:ptlrpc_server_handle_request()) Handled RPC pname:cluuid 
> +ref:pid:xid:nid:opc ldlm_cn_00:01318c63- 
> cfd4-9199-8142-4e41ea812bd3+7:32099:x9:12345-0 at lo:101
> 00000100:00100000:0:1216234366.071325:1536:32091:0:(ldlm_resource.c: 
> 719:ldlm_resource_add()) lvbo_init failed for resource 5820180: rc -2
> 00000100:00100000:0:1216234367.071325:1536:32091:0:(service.c: 
> 1064:ptlrpc_server_handle_request()) Handled RPC pname:cluuid 
> +ref:pid:xid:nid:opc ldlm_cn_00:01318c63- 
> cfd4-9199-8142-4e41ea812bd3+7:32099:x9:12345-0 at lo:101
>
>
> is PID 32091, from the client "0 at lo" (in this made up example a  
> local client).
> Then you can check on the client (also with "vfstrace" and "rpctrace"
> debugging on) what it was trying to do on the thread that requested  
> this
> RPC (PID 32091, XID "9" in this example).
>
> To quiet it, the easiest mechanism is probably to just delete this  
> file.
> If it is a small file (< 1MB) and the data is still valid you could  
> copy
> it to another file and rename it over the old one.
>
> Cheers, Andreas
> --
> Andreas Dilger
> Sr. Staff Engineer, Lustre Group
> Sun Microsystems of Canada, Inc.
>